Pathophysiology/Complications ORIGINAL ARTICLE

Overexpression of GLUT5 in Diabetic Muscle Is Reversed by Pioglitazone

CHARLES A. STUART, MD human , with mRNAs for MARY E.A. HOWELL, MS GLUT4, GLUT5, and GLUT12 predomi- DELING YIN, MD, PHD nating. Among these facilitative hexose transporters, GLUT5 is unique in that it transports only fructose (7). High dietary OBJECTIVE — This study was undertaken to quantify the expression of muscle GLUT in type fructose (but not ) content has 2 and to determine if treatment with an -enhancing thiazolidenedione drug, been shown to directly increase intestinal pioglitazone, would alter its expression. expression of Glut5 in a rat model (8). The current study demonstrated that mRNA RESEARCH DESIGN AND METHODS — Twelve patients with type 2 diabetes were concentrations for six of seven GLUTs in randomly assigned to treatment with either pioglitazone or placebo in a double-blinded 8-week diabetic muscle are not different from protocol. and mRNA for GLUT4 and GLUT5 were quantified in muscle homogenates from biopsies of vastus lateralis before and after treatment. The five additional GLUT family control subjects. Only GLUT5 mRNA dif- isoforms expressed in muscle had mRNA quantified in these samples. fered in the muscles from diabetic sub- jects. Both GLUT5 message and protein RESULTS — Baseline and posttreatment repeat measurements of GLUT4 protein were not were increased in baseline diabetic mus- different from control measurements. Compared with normal subjects, GLUT5 protein in- cle and decreased substantially with pio- creased 2.5-fold, and GLUT5 mRNA was 82% higher in the pretreatment samples from the glitazone treatment. diabetic subjects. Concentrations of mRNA for the six other GLUTs (GLUT1, GLUT3, GLUT4, GLUT8, GLUT11, and GLUT12) were not different from control subjects before or after treat- RESEARCH DESIGN AND ment. The proportion of type I (red) fibers (46%) in diabetic muscle was not affected by pioglitazone treatment. Pioglitazone treatment decreased muscle GLUT5 mRNA and protein by METHODS 52 and 40%, respectively, whereas placebo did not alter GLUT5 expression. Both red and white fibers had higher GLUT5 expression in the baseline diabetic muscle samples, and a pioglitazone- Materials related decrease in GLUT5 protein also occurred in both. Immunohistochemical studies were ex- amined using a Leica TCS SP2 Laser Scan- CONCLUSIONS — GLUT5 was dramatically increased in diabetic muscle, and pioglitazone ning Confocal Microscope (Wetzlar, treatment reversed this overexpression. The role of this fructose transporter expression in the Germany). An iCycler iQ System (Bio- insulin-enhancing effect of pioglitazone in muscle is unclear. Rad, Hercules, CA) was used for most of the PCR amplifications. The primers and Diabetes Care 30:925–931, 2007 standards for each isoform and glyceral- dehyde-3-phosphate dehydrogenase hiazolidenedione drugs facilitate in- are similar for these three drugs (GAPDH) were previously described (6). sulin action in patients with type 2 (1,4,5). The specific genes whose activa- Reverse transcriptase and oligo-dT prim- T diabetes manifested by improved tion or suppression are responsible for en- ers were purchased from Applied Biosys- glycemic control and a decrease in endog- hanced insulin action are unclear. The tems (Branchberg, NJ) as part of the enous insulin secretion (1,2). Drugs from study we describe here was designed to GeneAmp RNA PCR Core Kit. Affinity- this class robustly bind to the nuclear fac- determine if pioglitazone-induced en- purified rabbit anti-hGLUT5 (GT52-A) tor peroxisome proliferator–activator re- hanced insulin action on was purchased from Alpha Diagnostics ceptor-␥ and activate multiple could be associated with an increase in (San Antonio, TX). Mouse monoclonal cassettes (1,3). There are differences in expression of one or more GLUTs in skel- antibody specific for slow myosin heavy some of the genes activated by members etal muscle. We have previously shown chain (MAB1628) was purchased from of this class (troglitazone, rosiglitazone, (6) that seven members of the GLUT fam- Chemicon. AlexaFluor 647 donkey anti- and pioglitazone), but changes in many ily of hexose transporters are expressed in mouse, AlexaFluor 555 donkey anti- rabbit, and AlexaFluor 488 donkey ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● anti-goat were purchased from Molecular From the Department of Internal Medicine, East Tennessee State University, James H. Quillen College of Probes (Eugene, OR). NADH was pur- Medicine, Johnson City, Tennessee. chased from Sigma (St. Louis, MO). Address correspondence and reprint requests to Charles A. Stuart, MD, East Tennessee State University, James H. Quillen College of Medicine, P.O. Box 70622, Johnson City, TN 37614-0622. E-mail: SuperSignal west pico chemolumines- [email protected]. cence substrate was purchased from Received for publication 24 August 2006 and accepted in revised form 10 January 2007. Pierce (Rockford, IL). Published ahead of print at http://care.diabetesjournals.org on 24 January 2007. DOI: 10.2337/dc06- 1788. Abbreviations: AMPK, 5Ј-AMP–dependent protein kinase; GAPDH, glyceraldehyde-3-phosphate Patients and clinical protocol dehydrogenase. This protocol and the consent document A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion were approved by the East Tennessee factors for many substances. State University Institutional Review © 2007 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby Board. After informed consent was ob- marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. tained, 12 subjects were interviewed and

DIABETES CARE, VOLUME 30, NUMBER 4, APRIL 2007 925 Pioglitazone and GLUT5 in diabetic muscle

Table 1—Study subject characteristics

Diabetes Fasting duration insulin Subject Age (years) Sex (years) BMI (kg/m2) A1C (%) (pmol/l) Control subjects (mean Ϯ SD) 30 Ϯ 4.0 21.8 Ϯ 1.0 52 Ϯ 10.0 1 24 M — 21.8 — 29 2 26 F — 19.7 — 20 3 25 M — 23.4 — 74 4 51 F — 21.2 — 64 5 29 M — 25.5 — 44 6 26 F — 19.0 — 79 Placebo-treated subjects (mean Ϯ SD) 40.0 Ϯ 2.0 2.5 Ϯ 0.3 36.6 Ϯ 1.3 7.0 Ϯ 0.6 206 Ϯ 24.0 7 41 F 3 31.2 6.9 190 8 33 F 3 36.7 6.4 155 9 45 F 1 38.0 7.0 280 10 37 M 2 40.1 5.9 269 11 44 M 3 34.5 10.1 140 12 37 M 3 39.1 5.9 199 Pioglitazone-treated subjects (mean Ϯ SD) 46 Ϯ 5.0 3.7 Ϯ 1.5 32.1 Ϯ 2.2 7.3 Ϯ 1.0 166 Ϯ 34.0 13 47 M 1 25.5 5.9 207 14 55 F 6 37.1 8.0 232 15 53 M 2 34.0 7.2 143 16 24 F 10 37.7 11.7 266 17 49 F 2 25.8 5.3 70 18 45 M 1 32.4 5.6 77 examined. Six normal control subjects until the clinical protocol and all labora- least two separate times with triplicates of were also recruited. Control subjects were tory measurements were completed ex- each sample in each experiment. GAPDH nonobese with no diabetes in parents, cept for the GLUT5 immunoblotting. mRNA was quantified as an indicator of children, or siblings and were not being Western blots to compare GLUT5 protein RNA sample consistency. treated for an acute or chronic illness. Ta- levels were performed after it was deter- ble 1 displays the characteristics of con- mined that GLUT5 mRNA decreased in Immunoblot technique trol and diabetic subjects. In total, 10 of the pioglitazone-treated subjects. Immunoblotting was performed essen- the 12 subjects were obese with a mean Ϯ Either 1 week after screening or at the tially as previously described (10). For SE BMI for diabetic subjects of 34.3 Ϯ 1.4 end of the washout period, subjects were most GLUTs to be evaluated, 20 ␮g pro- kg/m2 (range 25.5–40.1). Subjects who randomized to pioglitazone treatment or tein from muscle homogenate was sepa- were taking insulin or who had diabetes placebo. Each subject underwent a percu- rated on a 10% polyacrylamide gel using Ͻ1 year were excluded. Three subjects taneous biopsy of vastus lateralis before the Laemmli system (11), transferred to a were being treated with sulfonylurea beginning medication and at the end of 8 nitrocellulose membrane, subjected to medications as monotherapy. Four sub- weeks of treatment. The muscle biopsy blocking with 2.5% nonfat dry milk in jects were being treated with was performed after an overnight fast and phosphate-buffered saline, incubated and a sulfonylurea, and two subjects were 2 h of quiet recumbency as previously de- with a validated dilution of one of the anti- treated with pioglitazone, one of whom scribed (9). Patients were instructed to GLUT antibodies above including 1.25% was also taking a sulfonylurea. Metformin maintain their prestudy activity and diet milk, and developed with enhanced and pioglitazone were discontinued, and programs throughout the 8-week study. chemiluminescence reagent and X-ray these subjects entered a 2-month washout Weekly phone contacts were made to as- film or a phosphorimager. period. The A1C level of the seven sub- sess any changes in home glucose moni- jects who went through the washout pe- toring. There were clinic visits at 4 weeks Immunohistochemistry riod did not change (mean 6.9 Ϯ 0.3% and again at 8 weeks, during which a sec- Confocal microscopic assessment of spe- before and 6.9 Ϯ 0.5% at the end). Sub- ond muscle biopsy was performed. At the cific fluorescent labeling of GLUT protein jects whose home blood glucose monitor- termination of the study, subjects were in normal human muscle sections were ing showed increased average values had returned to their prestudy regimen. performed using methods previously de- sulfonylurea medication added or in- scribed (6). Muscle fiber type composi- creased. At the end of the study, 8 of 12 Quantification of mRNA tion was determined by examining subjects were taking a sulfonylurea agent. All measurements of mRNA were per- sections stained with NADH oxidase as The study treatment protocol included formed using a Bio-Rad iCycler Thermal described by Scarpelli et al. (12) or by twice-daily 15 mg pioglitazone or a pla- Cycler with the iQ Real-Time PCR Detec- fluorescent staining using an anti-slow cebo pill. This study was double-blinded tion System as previously described (6). myosin heavy chain monoclonal antibody throughout, and the code was not broken GLUT quantification was performed at (6). All sections were coded, photo-

926 DIABETES CARE, VOLUME 30, NUMBER 4, APRIL 2007 Stuart, Howell, and Yin aaaemeans are Data 2— Table mRNA LT230 GLUT12 LT 185 GLUT4 LT 5.3 GLUT8 LT 56 GLUT5 L1 4.8 GLU11 LT 0.5 GLUT1 graphed, and then quantified indepen- mRNA to decline to within the normal 0.27 GLUT3 dently by three observers who were range. The GLUT5 mRNA data are shown unaware of which subject or treatment graphically in Fig. 1. Normalizing the the photograph represented. data to GAPDH mRNA concentrations diabetes 2 type with patients in transporters glucose for mRNAs muscle skeletal on treatment pioglitazone of Effect subjects† Ϯ

did not change the findings. The GLUT Control D * SD. Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Statistics isoform data we report here are raw data, Ϯ 923 2104 22 3209 63 . 7.8 2.1 . 6.1 0.9 . 1.7 0.2 All data are displayed as means Ϯ SE ex- not adjusted for GAPDH. 0.39 0.10 n cept Table 2, which lists SDs. Compari- ϭ sons between two groups were performed Change in GLUT5 protein related to † 5; n subjects‡

using Student’s t test for independent pioglitazone treatment Diabetic ϭ ;‡ 6; Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ groups, and when comparing paired mea- Immunoblots of muscle homogenates in- Ϯ .3 21 0.130 8 4009105 0.029 44 0044227 0.404 50 . .1 9.2 0.215 4.2 . .9 6.1 0.093 1.8 . .7 2.5 0.871 3.4 surements, the paired t test was used. dicated that subjects with type 2 diabetes 0.37 0.447 0.19 n

ANOVA was used for comparisons of who were not taking metformin or a thia- ϭ three independent groups. zolidenedione drug had increased RNA. nanogram per copies as expressed are data All 10. GLUT5 protein. Figure 2A shows a repre- RESULTS sentative immunoblot including control P subjects and baseline and posttreatment Impact of the 8-week protocol on muscle biopsies from a placebo-treated glycemic control subject and a pioglitazone-treated sub- weeks 8 Baseline Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ This protocol was double-blinded such ject. Individual data are displayed in Fig. Ϯ 719 442 24 4229 64 . 5.2 5.7 . 5.5 1.2 . 1.2 4.8 .60.40 0.16 that neither the subject nor the investiga- 2B. GLUT5 protein in the control subjects tors knew who was receiving active drug. averaged 100 Ϯ 3% of control, the base- treatment* Placebo treatment* Pioglitazone Nine subjects maintained their glucose line pioglitazone group 251 Ϯ 22%, the control during the study, but three sub- baseline placebo 255 Ϯ 31%, the pioglit- Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ jects in the placebo group had worsening azone posttreatment 151 Ϯ 16%, and the Ϯ 8 22 6 .6 192 0.963 2 165 2.5 3.0 1.6 of glycemic control, with A1C increasing placebo posttreatment 243 Ϯ 28%. 0.39 Pio- 0.996 0.03 0.32 by Ͼ1%. The mean A1C for the placebo glitazone treatment caused a 40% de- Ϯ group tended upwards (8.8 0.6 vs. crease. Baseline GLUT5 protein did not change Mean Ϫ Ϫ Ϫ Ϫ Ϯ ϭ Ϫ . .1 6.3 0.313 4.0 . .5 6.1 0.656 0.6 . .0 0.8 1.000 1.3 4005102 0.025 64 7.0 0.6%; P 0.07, paired t test) but correlate with age, weight, A1C, or fasting 25 0.742 2 did not change in the pioglitazone group plasma insulin concentration. (7.6 Ϯ 1.0 vs. 7.3 Ϯ 1.0%). Weight gain in excess of 1 kg occurred in five of the Skeletal muscle fiber type pioglitazone subjects and one of the pla- composition in patients with P cebo subjects. Weight change was 1.5 Ϯ diabetes 0.4 kg (P ϭ 0.008) for the pioglitazone Because GLUT5 is expressed predomi- group and Ϫ0.9 Ϯ 1.0 kg (P ϭ 0.47) for nantly in white muscle fibers, we quanti- weeks 8 Baseline Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 017 10 197 61 7197 27 . 6.2 1.6 . 6.8 2.4 . 0.8 0.9 the placebo group. Mean fasting plasma fied the proportion of fibers that were 0.63 0.25 insulin decreased slightly in both pioglit- type I oxidative (red) fibers in our subjects azone-treated (8%) and placebo-treated and determined whether that proportion (7%) groups, but was not statistically sig- was modified by pioglitazone treatment. nificant (P ϭ 0.15, P ϭ 0.39). However, Baseline biopsies had 47 Ϯ 3% (n ϭ 12) Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ plasma C-peptide decreased 21% (P ϭ red fibers in diabetic subjects compared Ϯ 54 11 4 0.625 5 140 2.6 . . 0.622 0.7 4.9 . 0.796 0 0.6 0.026) after pioglitazone treatment; the with 53 Ϯ 2% in sedentary control sub- 0.395 0.24 0.51 placebo-related drop of 6% was not sig- jects (n ϭ 6). Even though the proportion

ϭ change Ϫ Ϫ Ϫ

nificant (P 0.68). One muscle biopsy of red fibers was 13% lower in the diabetic Mean 0.847 5 .9 ( 6 0.191 8 was inadequate and one was lost, result- subjects compared with sedentary control 0.950 0.1 ing in one subject from each group not subjects, the proportion of white fibers being used in the muscle data analysis. was not altered by treatment. Postpioglit- Ϯ ϭ azone biopsies showed 45 2% (n 5) P Pioglitazone-induced changes in red fibers compared with 49 Ϯ 3% (n ϭ skeletal muscle GLUT mRNA 5) in the placebo-treated subjects. Fiber Ϫ ewe-ru oprsnof comparison Between-group Ϫ Ϫ ifrne(5 CI) (95% Difference Ϫ Seven GLUT mRNAs were quantified in composition alone could not account for ( 0.21 Ϫ . ( 3.9 . ( 1.3 9( 59 muscle homogenates from 6 normal con- the high amount of GLUT5 mRNA and 3( Ϫ Ϫ Ϫ Ϫ Ϫ trol subjects and from 10 patients with protein seen in the subjects with diabetes, Ϫ Ϫ 2t 4 0.437 24) to 12 2 o1)0.093 12) to 129 8 o18 0.980 178) to 182 20t .)0.297 4.1) to 12.0 .-t .)0.495 2.7) to 5.1- .6t .9 0.411 0.39) to 0.86 diabetes before and after an 8-week treat- nor could it explain the dramatic decline 1.000 1.3 ment with either 30 mg pioglitazone daily that occurred after treatment. change or placebo. Table 2 summarizes those data. Only GLUT5 mRNA was quantita- Muscle fiber type localization of tively different from the control subject pioglitazone-induced changes in data. GLUT5 mRNA was 82% higher in GLUT5 protein the baseline biopsy, and the treatment Muscle biopsies from each subject were P with pioglitazone caused the GLUT5 subjected to cryosectioning and immuno-

DIABETES CARE, VOLUME 30, NUMBER 4, APRIL 2007 927 Pioglitazone and GLUT5 in diabetic muscle

basolateral membrane of the intestinal lu- minal epithelial cell (15). GLUT5, GLUT7, GLUT9, and GLUT11 are all members of class II of the GLUT family, grouped together because of their high degree of amino acid identity (40–60%) and deduced to have fructose transport activity because of their similarity to GLUT5 (16). GLUT7 (17,18) and GLUT9 (19) subsequently have been confirmed to have fructose-inhibitable glucose up- take. In addition, GLUT8 and GLUT12 have been shown to transport both glu- cose and fructose (20).

Figure 1—Pioglitazone treatment alteration of GLUT5 mRNA in skeletal muscle from type 2 diabetic subjects. Shown are data from quantitative real-time PCR using a light cycler PCR instrument. Baseline GLUT5 mRNA increased 82% in subjects with diabetes, and pioglitazone treatment specifically decreased GLUT5 mRNA concentrations by 56%. The single asterisk (*) indicates significant difference from control subjects (P Ͻ 0.05, independent groups), and the double asterisk (**) represents significant difference from the pretreatment values (P Ͻ 0.05, paired observations). histochemistry. Figure 3A shows repre- In contrast, GLUT5 mRNA increased sentative sections. Comparing panels 1 82%, and GLUT5 protein increased more and 3 to panel 5 demonstrates that type I than twofold compared with control sub- and type II fibers of the diabetic subjects jects. Pioglitazone treatment of these pa- express more GLUT5 than the control tients for 8 weeks, while maintaining subject. Treatment with placebo had no glycemic control, caused a 52% decline in effect on the expression of GLUT5 pro- GLUT5 mRNA and a 40% decline in Figure 2—The impact of pioglitazone treat- ment on muscle GLUT5 expression in five sub- tein, but pioglitazone treatment resulted GLUT5 protein, but had no effect on jects with type 2 diabetes. A shows an in a substantial decrease in the GLUT5 mRNA concentrations for the remaining immunoblot of GLUT5 in muscle biopsy ho- protein content of both type I and II fi- GLUTs. Even though A1C level was un- mogenates from diabetic and control subjects. bers. Thereby, panel 4 qualitatively resem- changed, plasma C-peptide data sug- The GLUT5 protein band migrated with a rel- bles panel 5. This pattern and the change gested that endogenous insulin secretion ative molecular weight of 45 kDa. In this typi- or lack of change in fiber-specific expres- was modestly decreased coincident with cal immunoblot, lanes a, d, g, and h contained sion of GLUT5 was present in each of the pioglitazone treatment. 5 ␮g protein from homogenized muscle ob- 10 subjects with two evaluable biopsies. Glucose is by far the predominant tained by percutaneous biopsy from four nor- mal control subjects. Lanes b and c contained 5 Each biopsy of the 10 subjects treated in carbohydrate fuel used by mammalian ␮ this protocol was analyzed using image tissues, making it no surprise that 12 g homogenate from subject 4 before and after, respectively, 8 weeks of placebo treatment. analysis software to quantify the relative members of the 14-member GLUT family Lanes e and f contained 5 ␮g protein from mus- intensity of the GLUT5 immunofluores- of membrane transport glucose. cle homogenates obtained from subject 6 before cence. The results of these measurements Fructose, on the other hand, is found in and after 8 weeks of treatment with pioglita- are displayed in Fig. 3B. These data show blood at as much as 500-fold lower con- zone, 30 mg daily. B shows the individual data changes in fiber-specific GLUT5 protein centrations in humans (13). GLUT5, ex- points from each of the 10 completed subjects. expression that are similar to the changes clusively a fructose transporter with no Each data point represents the mean of deter- demonstrated in Western blots of muscle ability to transport glucose (7), has been minations from two to five separate experi- homogenates. demonstrated in intestinal epithelium, ments. GLUT5 expression was 2.5-fold higher kidney, fat, skeletal muscle, testes, and in diabetic subjects compared with control sub- CONCLUSIONS — Patients with sperm (14). Six other isoforms have been jects. Pioglitazone treatment resulted in a 40% decline in GLUT5 protein, whereas placebo type 2 diabetes had normal expression of shown to also transport fructose. GLUT2 had no effect. All of the pioglitazone-treated GLUT1, GLUT3, GLUT4, GLUT8, is a glucose and fructose transporter, subjects exhibited a decline in skeletal muscle GLUT11, and GLUT12 as indicated by which is likely responsible for fructose homogenate GLUT5 protein in the second bi- mRNA concentrations quantified using uptake in the liver (14) and plays a critical opsy, whereas only two of the placebo-treated reverse transcriptase and real-time PCR. role in glucose and fructose transfer at the subjects had a decline.

928 DIABETES CARE, VOLUME 30, NUMBER 4, APRIL 2007 Stuart, Howell, and Yin

Substrate availability has been shown to affect hexose transporter expression in intestinal epithelial cells but has not been directly evaluated in muscle. Hyperglyce- mia directly increases intestinal glucose absorption severalfold (21). Dyer et al. (21) demonstrated threefold increased transport of D-glucose into brush-border membrane vesicles isolated from duode- nal biopsies of patients with type 2 diabe- tes. They found that mRNAs for SGLT1, GLUT2, and GLUT5 were increased threefold in duodenal biopsies from these diabetic subjects. Protein levels for SGLT1 and GLUT5 were increased more than fourfold compared with control sub- jects (21). Dyer’s group hospitalized sev- eral patients with diabetes and intensively managed their blood glucose levels. They found that the overexpression of GLUT5 and SGLT1 in the intestinal epithelium returned to normal (21), suggesting that hyperglycemia had played a role in the overexpression of the transporters in the gut of the diabetic subjects. Experimental insulin-deficient diabe- tes is also associated with alteration in the expression of GLUTs. Streptozotocin- induced diabetes in rats results in dra- matic increases in both mRNA and

antibodies, as described in RESEARCH DESIGN AND METHODS. Panel 6 is the same as panel 5 except that it is probed with anti-slow myosin heavy- chain antibody to demonstrate which fibers are type I (red) fibers. B shows the change in GLUT5 expression induced by pioglitazone treatment in type I and II muscle fibers. Image analysis software was used to quantify the in- tensity of the red image due to anti-hGLUT5 primary antibody in sections from 6 control subjects and the 10 diabetic subjects before and after the 8-week treatment protocol. Images similar to those of Fig. 3 were used for each subject. The image analysis was performed us- ing before and after transverse muscle sections mounted on the same slide treated with the same antibody dilutions and imaged on the Leica confocal microscope with identical set- tings for all sections. The operator was not aware of which subjects received placebo or pioglitazone. Ten type I and 10 type II fibers were identified on each image, and the intensity of the GLUT5 signal was quantified using the Quantity One software from BioRad. The single Figure 3—Impact of pioglitazone treatment on the fiber type distribution of GLUT5 in muscle asterisk (*) indicates significant difference from patients with type 2 diabetes. Immunohistochemical methods were used to evaluate potential (P Ͻ 0.01) from the corresponding fiber type of change in fiber type distribution of GLUT5 in muscle biopsy material from patients treated with the control subjects. The double asterisk (**) pioglitazone. A displays representative immunohistochemical images. Panels 1 and 2 are from a indicates a significant treatment-induced dif- placebo-treated subject, and panels 3 and 4 are from a pioglitazone-treated subject. Panels 1 and ference (P Ͻ 0.01, repeated measures) from the 3 were from baseline biopsies, and panels 2 and 4 were from posttreatment biopsies. Panels 5 and corresponding fiber baseline data. Red fibers 6 are from biopsies of a normal subject. Cryosections from the two biopsies from each study subject (p) indicate type I, and white fibers indicate were placed on the same glass slide so that they were simultaneously incubated with the same type II, as indicated by slow myosin heavy- antibody solution. Each of panels 1–5 were probed with affinity-purified rabbit anti-hGLUT5 chain antibody fluorescence.

DIABETES CARE, VOLUME 30, NUMBER 4, APRIL 2007 929 Pioglitazone and GLUT5 in diabetic muscle protein for SGLT1, GLUT2, and GLUT5 same system that modulates the oxidative of mRNA for GLUT4 or GLUT12.We in jejunum and ileum (22). Insulin treat- enzyme content of muscle. Elite athletes speculate that the hyperglycemia of our ment of streptozotocin-induced diabetic may have as high as 75% type I (red and patients is accompanied by higher blood rats decreases the overexpression of these oxidative) muscle fibers (26), and sub- fructose levels that may directly stimulate GLUTs in intestinal epithelial cells (22). jects with type 2 diabetes have been GLUT5 expression in muscle by mecha- These data are consistent with an effect of shown to have as low as 35% type I with nisms that parallel the direct fructose ef- glucose concentration on GLUT expres- 65% type II (white and glycolytic) muscle fects on the intestinal epithelial cell. Our sion in the intestine. fibers (27). Type I fibers have higher lev- data show that enhanced insulin action In addition to endogenous hypergly- els of mitochondria and oxidative en- induced by pioglitazone is associated with cemia increasing intestinal glucose up- zymes and much higher levels of GLUT4 a major decline in the diabetes-related in- take, dietary content of sugars directly and GLUT12 (6) in contrast to type II fi- crease in GLUT5 expression in muscle. affects GLUT expression. Miyamoto et al. bers, which express low numbers of mi- The mechanism by which pioglitazone (8) fed rats a high-glucose diet and found tochondria but much higher levels of treatment decreases GLUT5 expression that intestinal uptake of glucose was in- GLUT5 (6). may be directly tied to its stimulatory creased coincident with increased jejunal Troglitazone, rosiglitazone, and pio- effects on mitochondrial oxidative SGLT1 and GLUT2 mRNAs. They found glitazone have each been shown to acti- enzymes. that D-glucose, D-galactose, and D- vate 5Ј-AMP–dependent protein kinase fructose feeding stimulated expression of (AMPK) through increasing the ratio of GLUT2, but only D-fructose could in- 5Ј-AMP to ATP in muscle cells (28,29). Acknowledgments— These studies were crease jejunal GLUT5 mRNA. Burant et al. Activation of AMPK increases the expres- funded, in part, by an investigator-initiated (23) showed that the D-fructose diet effect sion of peroxisome proliferator–activated grant from Takeda Pharmaceuticals. was due to direct contact of the sugar with receptor-␥ coactivator-1␣, the major reg- Special thanks go to Mary Ward, an out- standing research nurse who was essential to the intestinal epithelial cells and the effect ulator of mitochondrial biogenesis (30). completing the clinical arm of these studies. was rapidly reversible. These studies con- This action of thiazolidenedione drugs centrated on intestinal glucose transport shows important parallels to the effects of and did not evaluate the impact of glucose muscle contraction and exercise training or fructose on skeletal muscle expression on AMPK and peroxisome proliferator– References ␥ ␣ 1. Staels B, Fruchart JC: Therapeutic roles of of GLUTs. However, Darakhshan et al. activated receptor- coactivator-1 peroxisome proliferator–activated recep- (24) did not find an effect of fructose feed- (31,32). These effects are pro-oxidative in tor agonists. Diabetes 54:2460–2470, ing on fat or muscle expression of GLUT5. that they increase mitochondrial oxida- 2005 Kawasaki et al. (13) demonstrated tive enzymes and thereby increase ␤-oxi- 2. Miyazaki Y, Mahankali A, Matsuda M, that blood fructose concentrations were dation of fatty acids (33,34). Chronic Glass L, Mahankali S, Ferrannini E, Cusi elevated 50% and urine content was more stimulation of AMPK in rats has also been K, Mandarino LJ, DeFronzo RA: Im- than threefold increased in patients with shown to increase GLUT4 expression in proved glycemic control and enhanced diabetes. These investigators found that muscle (31). Taken together, these effects insulin sensitivity in type 2 diabetic sub- improving glycemic control resulted in on mitochondria and GLUT4 expression jects treated with pioglitazone. Diabetes normalization of the serum fructose con- would move diabetic muscle away from Care 24:710–719, 2001 3. Kersten S, Desvergne B, Wahli W: Roles of centrations and a marked decrease in the predominantly white, fast-twitch charac- PPARs in health and disease. Nature 405: urine excretion of fructose (13). teristics toward red, slow-twitch bio- 421–424, 2000 Our GLUT5 data from skeletal mus- chemistry. A drug treatment that acts 4. Hirakata M, Tozawa R, Imura Y, Sugiyama cle did not correlate with glycemic con- through gene expression to shift muscle Y: Comparison of the effects of pioglita- trol. The dramatic decrease in GLUT5 biochemical characteristics toward type I zone and rosiglitazone on macrophage protein and message in the pioglitazone- fibers may result in a reciprocal repres- foam cell formation. Biochem Biophys Res treated subjects was in spite of no signif- sion of GLUT5 gene expression. Commum 323:782–788, 2004 icant change in A1C. We believe that the high level of 5. Lee CH, Olson P, Evans RM: Minireview: The connection between GLUT5 and GLUT5 expression in type 2 diabetic lipid metabolism, metabolic diseases, and insulin action is unclear. GLUT5 is not muscle is directly tied to the predomi- peroxisome proliferator-activated recep- tors. Endocrinology 144:2201–2207, 2003 acutely regulated by insulin (25). The in- nance of type II muscle fibers within 6. Stuart CA, Yin D, Howell MEA, Dykes RJ, creased GLUT5 expression in both red which mitochondrial oxidative enzyme Laffan JJ, Ferrando AA: mRNA for hexose and white muscle fibers in type 2 diabetes expression is suppressed and GLUT5 ex- transporters GLUT4, GLUT5, and that decreases toward control subject lev- pression is increased. Pioglitazone treat- GLUT12 predominate in human muscle. els after treatment with pioglitazone sug- ment then reverses the gene regulation Am J Physiol Endocrinol Metab 291:E1067– gests an inverse relationship between cascades to shift muscle to a pattern of E1073, 2006 muscle GLUT5 and insulin sensitivity. gene expression that is more oxidative 7. Kane S, Seatter MJ, Gould GW: Func- The decrease in GLUT5 could be an indi- (red fiber–like), with increased mito- tional studies of human GLUT5: effect of rect or secondary effect of heightened in- chondria and decreased GLUT5. pH on substrate selection and an analysis sulin responsiveness, or it could be Our data show that pioglitazone- of substrate interactions. Biochem Biophys Rres Commum 238:503–505, 1997 somehow directly involved in the insulin- induced enhanced insulin action in dia- 8. Miyamoto K, Hase K, Takagi T, Fujii T, enhancing action of thiazolidinedione betic subjects is accompanied by Taketani Y, Minami H, Oka T, Nakabou Y: drugs. decreasing the high muscle expression of Differential responses of intestinal glucose Alternatively, pioglitazone may exert GLUT5 without any change in muscle fi- transporter mRNA transcripts to levels of its effects on GLUT5 expression via the ber type composition or in concentrations dietary sugars. Biochem J 295:211–215,

930 DIABETES CARE, VOLUME 30, NUMBER 4, APRIL 2007 Stuart, Howell, and Yin

1993 18. Manolescu A, Salas-Burgos AM, Fisch- Bjorntorp P: Muscle fiber composition 9. Carraro F, Stuart CA, Hartl WH, Rosen- barg J, Cheeseman CI: Identification of a and capillary density in women and men blatt J, Wolfe RR: Effect of exercise and hydrophobic residue as a key determinant with NIDDM. Diabetes Care 17:382–386, recovery on muscle protein synthesis in of fructose transport by the facilitative 1994 human subjects. Am J Physiol Endocrinol hexose transporter SLC2A7 (GLUT7). J 28. Fryer LG, Parbu-Patel A, Carling D: The Metab 259:E470–E476, 1990 Biol Chem 280:42978–42983, 2005 anti-diabetic drugs rosiglitazone and met- 10. Stuart CA, Wen G, Jiang J: GLUT3 protein 19. Doege H, Bocianski A, Scheepers A, Axer formin stimulate AMP-activated protein and mRNA in autopsy muscle specimens. H, Eckel J, Joost HG, Schurmann A: Char- kinase through distinct signaling path- Metabolism 48:876–880, 1999 acterization of human ways. J Biol Chem 277:25226–25232, 11. Laemmli UK: Cleavage of structural pro- (GLUT) 11 (encoded by SLC2A11), a 2002 teins during the assembly of the head of novel sugar-transport facilitator specifi- 29. LeBrasseur NK, Kelly M, Tsao TS, Farmer bacteriophage T4. Nature 227:680–685, cally expressed in heart and skeletal mus- SR, Saha AK, Ruderman NB, Tomas E: 1970 cle. Biochem J 359:443–449, 2001 Thiazolidinediones can rapidly activate 12. Scarpelli DG, Hess R, Pearse AG: The cy- 20. Rogers S, Chandler JD, Clarke AL, Petrou tochemical localization of oxidative en- S, Best JD: Glucose transporter GLUT12- AMP-activated protein kinase in mamma- zymes. I. Diphosphopyridine nucleotide functional characterization in Xenopus lian tissues. Am J Physiol Endocrinol Metab diaphorase and triphosphopyridine nu- laevis oocytes. Biochem Biophys Res Com- 291:E175–E181, 2006 cleotide diaphorase. J Biophys Biochem mum 308:422–426, 2003 30. Ruderman NB, Saha AK, Kraegen EW: Cytol 4:747–752, 1958 21. Dyer J, Wood IS, Palejwala A, Ellis A, Minireview: malonyl CoA, AMP-activated 13. Kawasaki T, Akanuma H, Yamanouchi T: Shirazi-Beechey SP: Expression of mono- protein kinase, and adiposity. Endocrinol- Increased fructose concentrations in saccharide transporters in intestine of dia- ogy 144:5166–5171, 2003 blood and urine in patients with diabetes. betic humans. Am J Physiol Gastrointest Liver 31. Winder WW, Holmes BF, Rubink DS, Diabetes Care 25:353–357, 2002 Physiol 282:G241–G248, 2002 Jensen EB, Chen M, Holloszy JO: Activa- 14. Bell GI, Burant CF, Takeda J, Gould GW: 22. Burant CF, Flink S, Depaoli AM, Chen J, tion of AMP-activated protein kinase in- Structure and function of mammalian fa- Lee WS, Hediger MA, Buse JB, Chang EB: creases mitochondrial enzymes in skeletal cilitative sugar transporters. J Biol Chem Small intestine hexose transport in exper- muscle. J Appl Physiol 88:2219–2226, 268:19161–19164, 1993 imental diabetes: increased transporter 2000 15. Helliwell PA, Richardson M, Affleck J, mRNA and protein expression in entero- 32. Suwa M, Nakano H, Kumagai S: Effects of Kellett GL: Stimulation of fructose trans- cytes. J Clin Invest 93:578–585, 1994 chronic AICAR treatment on fiber compo- port across the intestinal brush-border 23. Burant CF, Saxena M: Rapid reversible sition, enzyme activity, UCP3, and PGC-1 membrane by PMA is mediated by substrate regulation of fructose trans- in rat muscles. J Appl Physiol 95:960–968, GLUT2 and dynamically regulated by porter expression in rat small intestine 2003 protein kinase C. Biochem J 350:149–154, and kidney. Am J Physiol 267:G71–G79, 33. Wu H, Kanatous SB, Thurmond FA, Gal- 2000 1994 lardo T, Isotani E, Bassel-Duby R, Wil- 16. Joost HG, Thorens B: The extended 24. Darakhshan F, Hajduch E, Kristiansen S, liams RS: Regulation of mitochondrial GLUT-family of sugar/polyol transport fa- Richter EA, Hundal HS: Biochemical and biogenesis in skeletal muscle by CaMK. cilitators: nomenclature, sequence char- functional characterization of the GLUT5 acteristics, and potential function of its fructose transporter in rat skeletal muscle. Science 296:349–352, 2002 novel members (Review). Mol Membr Biol Biochem J 336:361–366, 1998 34. Koves TR, Li P, An J, Akimoto T, Slentz D, 18:247–256, 2001 25. Mueckler M: Facilitative glucose trans- Ilkayeva O, Dohm GL, Yan Z, Newgard 17. Li Q, Manolescu A, Ritzel M, Yao S, Slugoski porters. Eur J Biochem 219:713–725, 1994 CB, Muoio DM: Peroxisome proliferator- M, Young JD, Chen XZ, Cheeseman CI: 26. Bergh U, Thorstensson A, Sjodin B, activated receptor-gamma co-activator 1 Cloning and functional characterization Hulten B, Piehl K, Karlsson J: Maximal alpha-mediated metabolic remodeling of of the human GLUT7 isoform SLC2A7 oxygen uptake and muscle fiber types in skeletal myocytes mimics exercise train- from the small intestine. Am J Physiol Gas- trained and untrained humans. Med Sci ing and reverses lipid-induced mitochon- trointest Liver Physiol 287:G236–G242, Sports 10:151–154, 1978 drial inefficiency. J Biol Chem 280:33588– 2004 27. Marin P, Andersson B, Krotkiewski M, 33598, 2005

DIABETES CARE, VOLUME 30, NUMBER 4, APRIL 2007 931